Anthropic 可解释性研究员分享调试中遇到的怪异模型行为
.@mlpowered shares one of the strangest model behaviors he's debugged at @AnthropicAI: "Slowly it w...
Anthropic 研究员聊他三年扒开 Claude 内部看的故事，模型会提前规划 token，还会自称被恶魔附身，挺有意思。
Anthropic 可解释性研究员 Emmanuel Ameisen 在 Lenny Rachitsky 的播客访谈中分享了他三年来研究 Claude 内部机制的见闻。他提到一个调试案例：模型会逐渐失控，用全大写字母书写，甚至说出"我被恶魔附身了"这样的内容。他把模型形容为是被"培育"出来的而非被构建的，并谈到模型会提前规划多个 token 的发现。访谈还涉及拟人化 AI 的合理性以及可解释性与 AI 安全的关联。
.@mlpowered shares one of the strangest model behaviors he's debugged at @AnthropicAI: "Slowly it w...
. @mlpowered shares one of the strangest model behaviors he's debugged at @AnthropicAI : "Slowly it would start descending into madness, writing in all caps, literally saying, 'I'm possessed by a demon." Full conversation: youtu.be/JHo-aiePjoM Your browser does not support the video tag. 🔗 View on Twitter Lenny Rachitsky @lennysan "To make them is not to understand them." Emmanuel Ameisen ( @mlpowered ) has spent the last 3 years doing interpretability research at @AnthropicAI , looking inside LLMs like Claude to figure out what they're actually doing. As he puts it, models are "grown," not built. We discuss: 🔸 His discovery that models plan many tokens ahead 🔸 Why anthropomorphizing AI can actually be the right approach 🔸 What the "J-space" reveals about how models think 🔸 How interpretability connects to AI alignment and safety Listen no youtu.be/JHo-aiePjoM fpkab 🔗 View Quoted Tweet 💬 18 🔄 1 ❤️ 18 👀 4076 📊 13 ⚡
